Understanding Legal Meaning through Word Embeddings

For judges and legal scholars, the quest for meaning and the identification of methods appropriate for understanding word meaning animates volumes upon volumes of debate. Likewise, social scientists interested in studying the law have increasingly recognized the variations in choices over words as important barometers for understanding the law. In this paper, I suggest recent advances in computational linguistics -- notably, the efficient estimation of distributed representations of word meanings, or word embeddings -- offer a potentially transformative avenue by which to assess the strategy, choice, and impact of judicial language. Utilizing a corpus of more than one million federal and state appellate court decisions, I estimate word embeddings for the more than 400,000 most common words found in legal opinions. In a series of simple illustrations, I demonstrate the value of this treatment of word meaning for new avenues of law and politics research.